MORE: Merged opinions reputation model

Abstract

Reputation is generally defined as the opinion of a group on an aspect of a thing. This paper presents a reputation model that follows a probabilistic modeling of opinions based on three main concepts: (1) the value of an opinion decays with time, (2) the reputation of the opinion source impacts the reliability of the opinion, and (3) the certainty of the opinion impacts its weight with respect to other opinions. Furthermore, the model is flexible with its opinion sources: it may use explicit opinions or implicit opinions that can be extracted from agent behaviour in domains where explicit opinions are sparse. We illustrate the latter with an approach to extract opinions from behavioral information in the sports domain, focusing on football in particular. One of the uses of a reputation model is predicting behaviour.We take up the challenge of predicting the behavior of football teams in football matches, which we argue is a very interesting yet difficult approach for evaluating the model.